Print | Login
CIBM
Master thesis projects
anglais seulement CIBM > Student projects > Master thesis pro...

Navigate the CIBM


Homepage

Introduction to the CIBM

News

Events

Resources and Equipement

Volunteers

Positions

Student projects

   - TP IV projects
    - Master thesis projects

Publications

Where to find us

Available master thesis projects

BASP node project - Compressed sensing for high resolution functional magnetic resonance imaging

Master thesis project : available

The very new theory of compressed sensing (CS) generically aims at merging the two steps of signal acquisition and compression, surfing on the idea that a large variety of natural signals are sparse, i.e. that they can be expressed in terms of a small number of coefficients in some basis. It represents a significant evolution in sampling theory, beyond the well-known Nyquist-Shannon sampling theorem requiring a signal to be sampled at a frequency of twice its bandwidth to be exactly known. The theory demonstrates that for sparse signals a small number of measurements may suffice for an accurate and stable reconstruction. Considering a random incomplete Fourier coverage of a signal sparse in real space is a particular example of a good sensing procedure in this context. The theory only dates back to several years ago, but it already represents an active field of research and found many applications in various fields of science and technology.

Magnetic resonance imaging (MRI) represents one of the main techniques in use for biomedical imaging. Contrary to many alternative approaches, it is non-invasive and does not require ionizing radiation. It also offers a wide variety of modalities, ranging from structural to metabolic and functional imaging. Accelerating the MRI acquisition process is of major interest for many applications. Modern approaches seek to obtain the required reconstruction quality from a reduced amount of data, i.e. from an incomplete Fourier coverage. Given the sparse nature of the signals probed, compressed sensing was recently acknowledged as a powerful approach in this context.

Functional MRI (fMRI) aims at mapping the neural activity in the brain associated with specific tasks. The four-dimensional spatio-temporal signals under scrutiny are buried under both the background signal in the absence of activation, and measurement noise. The mapping of the activation requires a strong processing of the signal acquired, notably encompassing the analysis of the correlation between the time series pixel-wise and the known original task. In this extremely challenging context, the spatio-temporal resolutions currently achievable are often poor, so that any means for resolution enhancement would be of major interest. In this regard, the spatial spiky structure of the activation pattern is extremely sparse in image space, while its temporal structure is sparse in a domain of known waveforms corresponding to the task profile. In a compressed sensing approach this should allow for a drastic acceleration of the acquisition process, ultimately providing the resolution enhancement sought. The project will aim at studying such an approach, in comparison or conjunction with other acceleration techniques.

This research work will extend from theoretical and numerical grounds, to the acquisition and analysis of real data. It will be performed in collaboration with members of the BASP research node and of the Center for Biomedical Imaging (CIBM).

Requirements: Matlab/C programming, good knowledge of signal/image processing and/or biomedical imaging

Supervision :


BASP node project - Compressed sensing for accelerated diffusion magnetic resonance imaging

Master thesis project : available

The very new theory of compressed sensing (CS) generically aims at merging the two steps of signal acquisition and compression, surfing on the idea that a large variety of natural signals are sparse, i.e. that they can be expressed in terms of a small number of coefficients in some basis. It represents a significant evolution in sampling theory, beyond the well-known Nyquist-Shannon sampling theorem requiring a signal to be sampled at a frequency of twice its bandwidth to be exactly known. The theory demonstrates that for sparse signals a small number of measurements may suffice for an accurate and stable reconstruction. Considering a random incomplete Fourier coverage of a signal sparse in real space is a particular example of a good sensing procedure in this context. The theory only dates back to several years ago, but it already represents an active field of research and found many applications in various fields of science and technology.

Magnetic resonance imaging (MRI) represents one of the main techniques in use for biomedical imaging. Contrary to many alternative approaches, it is non-invasive and does not require ionizing radiation. It also offers a wide variety of modalities, ranging from structural to metabolic and functional imaging. Accelerating the MRI acquisition process is of major interest for many applications. Modern approaches seek to obtain the required reconstruction quality from a reduced amount of data, i.e. from an incomplete Fourier coverage. Given the sparse nature of the signals probed, compressed sensing was recently acknowledged as a powerful approach in this context.

In diffusion MRI (dMRI), multiple three-dimensional signals are probed, each of them associated with molecular diffusion in a specific spatial direction. In other words, a signal on the sphere is probed in each voxel, hence defining an orientation distribution function. A precise mapping of this orientation distribution function is notably of fundamental interest for tractography of the neuronal connectivity in the brain. In this context, incremental signals, simply defined by the difference between signals associated with the diffusion in different directions, may appear to be sparser than each signal separately. In a compressed sensing approach this would allow further acceleration of the acquisition process than what is reachable for each signal separately. The project will aim at developing corresponding reconstruction algorithms and studying the practical acceleration that such an approach can provide, in comparison or conjunction with other acceleration techniques.

This research work will extend from theoretical and numerical grounds, to the acquisition and analysis of real data. It will be performed in collaboration with members of the BASP research node and of the Center for Biomedical Imaging (CIBM).

Requirements: Matlab/C programming, good knowledge of signal/image processing and/or biomedical imaging

Supervision :


In vivo 13C NMR spectroscopy in the rat brain

Master thesis project : available

Non-invasive techniques were developed to image brain activity such as fMRI and PET scanning. Although routinely used, these techniques measure some features that are consequence of neuronal activity, including hemodynamic responses, glucose uptake, glucose phosphorylation or oxygen consumption. The brain is compartmentalized in mainly neurons and glia that maintain intrinsic relations and cooperate to fuel the electrical activity and neurotransmission. 13C NMR spectroscopy has been strategically used to probe the neurochemical and metabolic interactions between brain compartments and to investigate neuroenergetics of brain function.
The purpose of this project is to simultaneous measure oxidative metabolism namely glycolysis and the tricarboxylic acid cycle, neuron-glia interaction through the glutamate-glutamine cycle, anaplerosis and the malatate-aspartate shuttle. The high sensitivity achieved with the use of high-fild magnets and the use of different specifically 13C-enriched substrates may allow unraveling some controversial features of cerebral metabolism and its relation to brain function.

The project will involve not only NMR spectroscopy studies but also the employment of mathematical models of cerebral metabolism and it is suitable for students with a background on biomedical physics, biochemistry or biology who are interested in neurochemistry.

Supervision :


Cerebral blood flow of mice at 14.1T

Master thesis project : available

We've established a two-coil system for continuous arterial spin labeling (CASL) at 9.4T. This allowed us studying animal disease models and help understanding underneath causes for dysfunctional blood flow in brain. It's known that increased magnetic fields amplify sensitivities and shorten acquisition times, and thus help studying transgenic modified mice.

This project is suitable for students with a background in Physics or Medical Physics or Biomedical Electrical Engineer with an interest in studying magnetic resonance instrumentations or electronics. The aim of study is to construct a imaging coil, one of the components of the two-coil system for CASL at 14.1T. Work will involve mechanical and electrical improvements of the coils, soldering the parts together, testing the coil at the bench using a network analyzer and at the scanner. If time allowed, the MR results will be compared with those acquired at 9.4T.

Supervision :


Construction of a 14T volume coil

Master thesis project : available

At CIBM we have two high-field animal MRI/S systems. Currently we use surface coils for both imaging and spectroscopy, giving high sensitivity over a small volume. For some applications, such as fMRI or MEMRI, it is more desirable to have homogeneous sensitivity over the whole animal than high sensitivity over a small part of it. For this we need a volume coil.

The aim of this project is to design and build a Birdcage coil for imaging rat brains using the 14T scanner (600MHz). Work will involve mechanical and electrical design of the coil, soldering the parts together, testing the coil at the bench using a network analyser, testing the coil at the scanner using a phantom and finally acquiring images of a rat. If time permits, the coil could be made active-detunable, to enable it to be used with a separate surface receive coil.

This project is suitable for students with a background in biomedical physics or physics with an interest in electronics. You will gain experience in radiofrequency design and testing.

Supervision :


Improving image anatomical and functional image quality via parallel transmission

Master thesis project : available

The short wavelength used in RF pulses to excite protons at high magnetic fields (such as our 7T MR system), which are smaller than the imaged object (in our case, a human head), generate artifacts in the images due to constructive and destructive interference of these RF waves. Parallel transmission offers a very promising approach to correct for this radiofrequency inhomogeneities. We have recently acquired a system capable of performing parallel transmission and this project would involve creating protocols for B1 mapping, followed by the implementation in MATLAB of code to calculate the B1 maps of each coil and the coil combination that would allow creating the smoothest final image. In its simplest implementation two gradient echo images with two different flip angles are acquired and the resulting image intensities are used to calculate the spatial flip angle distribution of each separate coil.
Subsequently the aim is to test different algorithms that allow calculation of how the different coils should be combined to achieve a spatially homogeneous excitation.
As a conclusion of the project, sequences that have a significant dependency on the excitation flip angle will be tested to demonstrate the improvements obtained with parallel transmission.

Supervision :


BOLD fMRI of the mouse barrel cortex

Master thesis project : available

Functional magnetic resonance imaging (fMRI) is an important tool for the study of the function, plasticity and connectivity in the rodent brain under normal and pathological conditions, as well as for the interpretation of the BOLD response relative to neural activity. Mouse fMRI remains a challenging task due to the small size of these animals and the difficulty to maintain their physiological condition during stimulation.

The purpose of this project is to investigate and validate the BOLD response in the mouse barrel cortex at 9.4T or 14.1T using a variety of stimulation paradigms. The project will involve the development of an appropriate setup for fast imaging of mice at high field. Other techniques such as optical imaging and electrophysiology will be used to validate our experiments.

The project is suitable for students with a background in physics, biomedical physics or biology who are interested in biomedical applications of magnetic resonance imaging.

Supervision :


Motion Detection using Navigators

Master thesis project : available

Magnetic resonance imaging (MRI) suffers from its inherent susceptibility to patient motion. As a consequent, a significant amount of acquired data cannot be used for diagnosis in clinical practice, which leads to patient inconvenience and considerable costs.

One way to overcome this problem is to introduce a so-called "navigator", which obtains a small amount of additional data during the image acquisition process. These data is then used to assess if the patient has moved, which can be further on used to correct the images. The proposed project would encompass to investigate navigator data acquired with different imaging and spectroscopic sequences and to develop methods to detect motion using these data. The student would also be involved in acquiring the necessary data sets from healthy volunteers.

This project is suitable for students of bio-or biomedical physics, physics, and instrumentation. Matlab knowledge is highly appreciated.

Supervision :


Optimization of the Turbo Spin Echo sequence for 7T applications

Master thesis project : available

Turbo Spin Echo sequences have proven to be a very useful in the clinical environment (1.5 and 3T) as they provide high SNR T2 weighted images with a short total acquisition time. At high magnetic fields (>7T), because of its high inherent SAR and the inhomogeneity of the transmission field, turbo spin echo sequences haven't been broadly explored. In this project we seek to optimize flip angle amplitudes in the context of hyperechos and TRAPS, to reduce SAR so that whole brain high resolution images could be acquired under 10mins. The project will involve numerical calculations of the magnetization evolution throughout the sequence, optimization of the contrast between white and grey matter, while reducing the SAR. The results of this optimization will be evaluated with in vivo data to be acquired at our 7T system.

Supervision :


Localized in vivo 1H and 15N MRS in the hyperammonaemic rat brain

Master thesis project : available

An alternative approach to 13C MRS for studying glutamate and glutamine metabolism is 15N MRS during infusion of 15N labeled ammonia. Ammonia is metabolized to glutamine by glutamine synthetase (GS) in astrocytes. Consequently, the activity of cerebral glutamine synthetase in vivo under hyperammonaemia conditions may help understand the mechanism of ammonia toxicity and could provide further insight into the glutamate-glutamine cycle.
All the in vivo data will be acquired using a 9.4T animal MR scanner. In vivo localized 15N MRS will be used interleaved with in vivo 1H MRS to measure the glutamine synthesis rate under 15N labeled ammonia infusion in the rat brain.

This project is suitable for students with a background in biochemistry, biomedical physics or physics.

Supervision :


Vessel size index magnetic resonance imaging at high field

Master thesis project : available

fMRI, Positron emission tomography and NIRS all rely on tight coupling between focal cerebral hemodynamics and neuronal activity. A detailed understanding of the mechanisms of local cerebral circulatory regulation is thus critical for the establishment of accuracy of these techniques.

The purpose of the present work is to investigate the changes in cortical vessel diameter by inducing mild hypercapnia. Changes in T2 and T2* following injection of a high molecular weight superparamagnetic contrast agent (SINEREM, Guerbet) are measured using a dedicated MR imaging sequence. The ratio of ΔR2 and ΔR2* changes gives an estimation of the vessel radius according to microvasculature models described in the literature. This project will involve the development of an EPI-based sequence at 9.4T/14T, its evaluation for the measurement of T2 and T2* in the rat/mouse brain under normocapnic and hypercapnic conditions.

Supervision :


Integration of B1+ Field (Flip Angle) Mapping into the Reconstruction Software of Siemens High Field MRI Scanners

Master thesis project : available

To assess the homogeneity of the radiofrequency (RF) excitation field B1+, flip angle mapping is often performed in nuclear magnetic resonance (NMR) applications. Due to the shortened wavelength at higher frequencies, B1+ field mapping has gained increased importance for high field (>=3Tesla (T)) magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS). In its simplest implementation two gradient echo images with two different flip angles are acquired, and the resulting image intensities are used to calculate the spatial flip angle distribution.
In this project, code for the B1+ (or flip angle) mapping has to be integrated into the reconstruction software of two state-of-the-art Siemens high field MRI systems (3T and 7T). The goal is to be able to perform measurement, data analysis, and display of the maps directly on the scanner.
Strong C++ skills will be of great advantage, since programming will have to be carried out in the C++ language.

Supervision :


Short-echo-time 1H MRS of the mouse lacking prion protein at 14.1T

Master thesis project : available

The prion diseases form a group of fatal neurodegenerative diseases, also described as transmissible spongiform encephalopathies (TSEs), which are caused by abnormal conformational isomers (PrPSc) of the host-encoded prion proteins (PrPc). PrPc is expressed in all vertebrate species examined to date, mainly in the brain but also in many other tissues at lower levels. The nature of the prion has been a longstanding enigma. The mechanism by which prions elicit brain damage and the relative contributions of PrPSc accumulation and PrPc depletion to the prion replication remains unclear. Consequently, different animal models were created in order to study the role of the PrPc. Among these models, knockout mouse models were crucial in elucidating the precursor-product relationship between PrPc and PrPSc. On histologic analysis of tissue, spongiform degeneration and astrocytic gliosis was observed in mice lacking of prion protein. In vivo 1H MRS (proton magnetic resonance spectroscopy) allows non invasive characterization of brain metabolism and it has been used to study brain metabolites changes in a wide range of neurodegenerative diseases. Only a very few reports describing the use of in vivo 1H MRS in prion disease have been published so far, reporting only a few metabolites. From our knowledge, no in vivo MRS study was performed until now in mice lacking prion protein.

The purpose of this study is to use in vivo high-resolution 1H MRS at 14.1T to measure the neurochemical profile (also called brain metabolites) in mice lacking prion protein with ultimate goal of a rapid and precise determination of the neurochemical changes.

The project will involve data acquisition (imaging and localized spectroscopy) at 14.1T and data processing (determination of brain metabolite concentrations).

This project is suitable for students with a background in physics or biomedical physics, who are interested in biomedical applications of proton magnetic resonance spectroscopy.

Supervision :


A study of treatment of transient ischemia in mouse brain using proton high-resolution spectroscopic imaging

Master thesis project : available

This project is suitable for students with a background in MR physics, who are interested in biomedical applications of magnetic resonance.

Cerebral stroke is one of the most frequent causes of disability and death in humans. However, less affected brain tissue in early stages of stroke can be recovered, especially as a result of a proper drug treatment. Metabolite changes in the stroke region of mouse brain will be studied by high-spatial-resolution proton MR spectroscopy on a unique 14 Tesla animal MR scanner.

A protocol for high-resolution proton spectroscopic imaging of mouse brain has already been developed. However, the measurement time is still too long for practical applications. Thus, optimization of the experimental protocol, possibly combined with the use of a new array RF coil, is necessary. The obtained data will be processed with a newly developed data processing tool (written in MATLAB). Metabolite maps of the ischemic mouse brain, without and with treatment, will be obtained and the effect of the treatment will be evaluated.

Supervision :


Development of biomedical instruments to measure highly polarized nuclei in MRI

Master thesis project : available


Hyperpolarized Magnetic Resonance is a revolutionary and promising technique with a proven potential in biomedicine and biochemistry. Based on Dynamic Nuclear Polarization (DNP), hyperpolarization leads to an improvement in signal-to-noise ratio on the order of 10'000 times. It allows for the detection of tiny amounts of biological substances in vivo and to follow the metabolic pathways of chosen molecules.

Two unique hyperpolarizers have been developed within the Swiss DNP Initiative regrouping several departments of the EPFL and the Paul Scherrer Institute. The Center for Biomedical Imaging (CIBM) provides access to two high-field animal MRI systems, one working at 9.4T and the other at 14.1T. One of the hyperpolarizers is currently coupled to the 9.4T scanner. For further information on our Initiative, see the Swiss DNP Initiative website.

The aim of the proposed project is to develop a tool and a protocol to measure the polarization of the nuclear spins just before the infusion of hyperpolarized solution into the animal. This will greatly help the analysis of the in vivo data. The project will involve the development of radio-frequency circuits that can be adapted to the infusion devices located inside the MRI scanner. The setup will be tested in real experiments with hyperpolarized solutions using world-class magnetic resonance instrumentation.

Supervision :


A biocompatible coil for 1H MRS of rodent at 9.4-T

Master thesis project : available

Goal of study: To design, construct and test a biocompatible coil resonating at 400 MHz

In vivo Magnetic Resonance Spectroscopy (MRS) allows non-invasively assessing metabolite concentrations of targeting tissue, such as brain. However, the sensitivity of MRS is reduced due to the performance of coils, one of the key components in MR, which normally are located some distance from the region of interest (ROI). Intrinsically, minimizing the distance between the probe (coil) and the ROI would improve the quality of data, for example, the sensitivity with less power. One possible solution would be embedding coils with a biocompatibility feature, which allows those coils to be compatible with living tissue or a living system by not being toxic, injurious, or physiologically reactive and not causing immunological rejection. Therefore, the coils could be placed with a minimal distance to the ROI. The addition challenges are the relative ease and reproducibility in their positioning in order to target specific, localized regions under.

The purpose of this project is to design and construct such a coil and evaluate its function at 400 MHz for in vivo 1H MRS. This project has high biomedical priority, not only for its originality in the field but also possible long-term studies (implantable) under physiological and pathophysiological conditions.

The project will involve understanding the principles of coil design for MRI and MRS.

Duration of project: 4 to 6 months

Supervision :


Investigating the influence of MRI acquisition parameters on brain tissue modeling and statistical classification

Master thesis project : available

Accurate and robust brain tissue segmentation from MR images is a key issue in many applications of medical image analysis for quantitative studies and particularly in the study of neurodegenerative disorders such as Alzheimer's disease. Statistical unsupervised approaches have been developed to automatically classify the brain into white matter, gray matter and cerebrospinal fluid. These methods rely on hypotheses regarding the intensity and the spatial models. However, these assumptions may become invalid if MR sequence parameters are not appropriately chosen.

The goal of this project is to assess the accuracy and robustness of tissue models and unsupervised classification methods with regard to the T1-weighted MRI acquisition parameters. As a result, optimal criteria will be determined and integrated in a "quality check" tool that will be implemented in Matlab and/or C++.

Supervision :


Site map| June 2005
La responsabilité éditoriale de cette page est assumée par son auteur